I want to make an electret microphone setup for my accordion. 4 mics on the treble side and 2 on the bass side.

Would I need a preamp for each mic and then mix the outputs using an opamp?

Or would it be possible to make it not using preamps or an active mixer? Would the signal be loud enough for a sound card to then amplify for example?

It would be nice if the mics could be powered by 48v phantom power.

Could this circuit work? Using preamplifiers and an opamp mixer enter image description here

  • 1
    \$\begingroup\$ Two microphones close-ish together on the treble side will, inevitably produce signals on each microphone that destructively cancel out. Have you properly thought this idea through? \$\endgroup\$
    – Andy aka
    Commented Dec 7, 2021 at 16:34
  • \$\begingroup\$ I just know that most commercial microphone setups for accordions use multiple microphones close to each other. \$\endgroup\$
    – elt
    Commented Dec 7, 2021 at 16:36
  • \$\begingroup\$ At the very least, you're going to want to allow for the possibility of a need for different gains on each mic. \$\endgroup\$ Commented Dec 7, 2021 at 16:48

5 Answers 5


I did a microphone bar for our accordion orchestra and it was way simpler than what you envision. I used a mini-XLR plug like those used on wireless transmitters: AKG apparently uses 4V through 4kOhm. Found that out later: my amp now has gotten 3.7V through 1.2kOhm instead. The 4V through 4kOhm variant does work as well, though. As does a +48V to mini-XLR adapter that comes with a number of affordable instrument microphones.

The circuit for the microphone capsules just put 4 capsules in parallel. For volume control I ended up with a 5kOhm linear pot (you really want to use this for fine adjustments and not as a general volume control so a log pot gets too fiddly) with the wiper connected to the combined "hot" from the mics (doing it the other way round gives you wiper noise of constant volume rather than wiper noise proportional to the loudness, quite irritating at lower levels), the low side of the mics connected to signal ground, the low side of the pot connected to signal ground through a 47uF Tantal capacitor (this makes sure that the mics receive full voltage even at low volume settings) and the high side of the volume pot going to hot+power (pins 2+3) in the mini-XLR. Somewhat paranoically I wired the microphone bar exterior (electrically separated) to shield in the connecting cable and mini-XLR and put signal ground and combined hot+power through the two inner lines of a thin microphone cable. I used a 3.5mm 4-way socket in the bar itself, putting hot+power on the tip and first ring, signal ground on the second ring, and shield on shield.

That way I can connect the microphone bar in a pinch to video microphone inputs with a simple 3.5mm TRS cable instead of the mini-XLR ended cable.

I didn't give the 4 mics quite the same distance in order to avoid stacking comb filtering effects. If you do the shielding carefully and have a noise-free power-supply (the voltage reference of a voltage regulator is too noisy without additional capacitors), you don't need active components apart from the mics themselves to get reasonably noise-free signal, assuming capsules of good size. At least for live play (we use this for a bass accordion and after I was through with the amp as well, there is no background noise or hum between playing).

Bar itself is an aluminum U profile with rubber lining on the edges to block outside sound well enough to avoid feedback. Buttonhole elastics are used for attaching to the grille screws. Padded textile tape (for bicycle bars) reduces microphoning from the bar itself.

Have fun finding your own solutions!


It would pick up audio from all mics in parallel sharing the drain (electret+) pullup R that may need to be adjusted to get full range at V+/2. (phantom DC power to 12V Zener thus 6V is ideal on electret for gain and max swing.) (not necessary, but possible)

This is a balanced electret version with +=drain and - source between two matched 2k2 resistors.

However, amplitude phase may not be coherent in higher range of f and can create nulls when amplitudes are equal and out of phase or add amplitudes when in phase. This is also partly due to the amount of open rear air filter designed to attenuate distance audio pressure yet be more sensitive to front pressure from direct line to source. Walls can also induce echos or cancellation for lower frequencies to a lesser extent. First determine how much far field attenuation you get with a distant speaker swept using Audacity FM or sweep generator and record and compare all mics. Some are very well-balanced, so the rear cancels the front by 40 dB which is good for blocking ambient noise in the distant and preventing feedback from speaker monitors. Conversely, blocking the rear inlet makes them very sensitive to distance audio, which is good for conference type calls with a room of people.

Electret mics are not high quality and differences in performance must be tested before installation using a sweep from a good tweeter in a sound baffled room then measure the response of each then paired using audacity to see if my assumptions are valid or not for your application.

But by careful testing with room dynamics and a good speaker, Audacity and different mounting locations, testing each location for differences in the "spectral response plot" then re-testing with iterations, and considering different R2 values, you may find the right attenuation needed to prevent clipping and amplitude/phase response without phasing errors.


I cannot verify if your schematic is optimal for multiple electrets.

Avoiding RF noise and grid hum is also important. Read

For minimum hum pickup the two 2k2 bias resistors should be perfectly matched.


U1B, the summing opamp, is missing negative feedback to set the circuit gain.

Fir the input volume control issue, add a fixed resistor between each pot wiper and the corresponding inverting input. This resistor will set the maximum voltage gain of the preamp.

If you have a maximum gain value in mind for the overall signal path through two opamps, split it between gain in the mic preamp and gain in the summer.


EEK! Your schematic is an impossible to see NEGATIVE image. I made part of it a normal POSITIVE image. Your preamp circuit will not work an an inverting opamp with a volume control. Use a non-inverting opamp circuit instead.


EDIT: I notice that the opamp inverting input is grounded through the volume control but the non-inverting input is at half the power supply voltage. Then the output of the opamp will be as high as it can go and it will not amplify the audio.


Electret capsules in two-wire mode can be just put in parallel since they are essentially variable current sources.

However, accordions are loud. You need comparatively insensitive capsules to avoid clipping, or you need to operate on the capsules to convert them to 3-wire mode. In that case, they stop being current sources, so putting them in parallel becomes a trickier proposition.

However, you can still maintain them as current sources while lowering their sensitivity by cutting the correct trace on their bottom (search for "Linkwitz modification" for instructions of where to cut) and inserting a resistor between case and the JFET's source pin.

A typical electret condenser mic JFET is K596 with a typical transmittance T of 1.2mS. After inserting a resistor R at the source, the resulting transmittance is \$\frac{1}{1+RT}\$. So you can halve the sensitivity by putting in 830ohms and so on. Of course this involves a DC voltage loss across the resistor, so you may need to adapt your operating voltage.

If you are using opamps for amplification, you can of course sum signals via resistors on a negative opamp input as virtual ground, using inverting amplifier configuration.

But you might be fine without any additional active components.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.